Comparative transcriptomics analysis of histone deacetylases, transcription factors, and ion channel genes in human iPSC-cardiomyocytes vs. the adult human heart

Figure 1.. Epigenetic de-acetylation modifiers and cardiac ion channels: experimental design to compare their role in human iPSC-CMs and adult human cardiac tissue. A. HATs and HDACs epigenetically regulate gene expression through reversible (de)acetylation of histone proteins and through direct action on TFs with potential regulatory effects on cardiomyocyte electrophysiology. In addition to catalytic action on chromatin compaction/relaxation, the HDACs and HATs have also non-canonical roles as protein stabilizers, chaperons, metabolic modulators, regulators of viral entry. B. These relationships were modelled computationally with HDAC/SIRT and HAT expression as inputs, cardiac ion channel expression as outputs, and TF expression as intermediaries. C. Cellular localization of HDACs and SIRTs corresponds to roles in chromatin remodeling, protein interaction, mitochondrial regulation, etc. Different classes of HDACs and SIRTs are color-coded. D. Experimental design and timeline for siRNA treatment includes RNA sequencing of hiPSC-CMs, which was processed alongside RNAseq data (adult human heart) accessed from the GTEx database. A separate set of 48 h siRNA treated live hiPSC-CM samples were…

Figure 2.. Comparative profiles of gene expression of cardiac ion channel genes in hiPSC-CMs and in human adult left ventricle. (A) Relative expression of select cardiac ion channel genes and (B) more comprehensive cardiac electrophysiology “rhythmonome” genes in hiPSC-CM (n=61 female) and adult LV (n=84 female, from GTEx) bulk RNA-sequencing data. The inset in panel A is a radial plot of the expression averages for the two groups to highlight pattern similarity of expression. Kdifference is listed as well. Individual data points displayed; error bars represent mean and SEM.

Figure 3.. Comparative profiles of gene expression of HDACs/SIRTs, HATs, and TFs in hiPSC-CMs and in human adult left ventricle. Relative gene expression of HDACs/SIRTs (A), HATs (B), and key cardiac transcription factor genes (C) in hiPSC-CM (n=61) and adult LV (n=84, from GTEx) bulk RNA-sequencing data. The radial plots of the expression averages for the two groups are shown to the right of panels A, B and C to highlight pattern similarity or differences of expression. Kdifference is listed for each gene set. Individual data points displayed; error bars represent mean and SEM.

Figure 4.. Whole-transcriptome (RNA sequencing) UMAP clustering of hiPSC-CMs treated by siRNAs against HDACs. (A) Results for targeting HDAC classes I, II, and IV (B) Results for targeting HDAC class III. UMAP embedding performed after univariate data imputation and scaling. Samples color-coded by HDAC class. N1 and N2 are negative controls treated by scramble siRNA on Plate 1 and Plate 2 respectively. U1 and U2 are untreated controls on Plate 1 and Plate 2 respectively. Dotted lines mark samples that clustered relatively tight and far away from the control groups. For U1 group, n=4; for all other treatment groups, n=3. Bi-clustering heatmaps display log2 transformed expression values, sorted by their adjusted p-value for top 30 differentially expressed genes between the HDAC9 knockdown group and corresponding negative control (C) and between the SIRT3 knockdown group and corresponding negative control (D). Volcano plots display global transcriptional change across HDAC9 knockdown group and negative control (E) and across SIRT3 knockdown group and negative control (F). Each dot represents one gene. Genes with adjusted p-value < 0.05 and log2 fold change > 1 are colored red and are…

Figure 5.. Analysis of relationships between HDAC/SIRT genes and ion channel genes in hiPSC-CM and adult LV bulk RNA-sequencing data. A. Correlation matrices where positive/negative correlations are colored green/red respectively. B. Results from PLS regression models are presented as biplots for the model inputs and outputs (HDACs/SIRTs and ion channels, PLS1) for hiPSC-CMs and for adult LV. Data for adult LV are modified from Pressler et al.(Pressler et al., 2022). C. Average relative feature importance of HDAC/SIRT inputs in predicting ion channel outputs are displayed for hiPSC-CM and adult LV PLS models.

Figure 6.. Analysis of relationships between HDAC/SIRT genes and TF genes in hiPSC-CM and adult LV bulk RNA-sequencing data. A. Correlation matrices show positive/negative correlations colored in green/red respectively. B. Partial least squares regression models are presented as biplots for the model inputs and outputs (HDACs/SIRTs and TFs, PLS2). Data for adult LV are modified from Pressler et al. (Pressler et al., 2022).

Figure 7.. Analysis of relationships between TF genes and ion channel genes in hiPSC-CM and adult LV bulk RNA-sequencing data. A. Correlation matrices show positive/negative correlations colored in green/red respectively. B. Partial least squares regression models are presented as biplots for the model inputs and outputs (TFs and ion channels, PLS3). Data for adult LV are modified from Pressler et al.(Pressler et al., 2022).

Figure 8.. Links between HDACs/SIRTS, TFs and cardiac ion channels. Integrated (Sankey) visualizations depict links between HDACs/SIRTs (left), TFs (middle), and ion channels (right) in hiPSC-CM (A) and adult LV (B). Color and weight of lines defined by sign and magnitude of the PLS β-matrix coefficients.

Figure 9.. Modulation of excitability and spontaneous rates of hiPSC-CMs by suppression of HDACs and SIRTs. A. Spontaneous beating rates of hiPSC-CM treated for 48h with siRNAs against individual HDACs and SIRTs, displayed as percent change from the negative siRNA control. Isoproterenol and propranolol were included as positive and negative rate controls respectively, although after 48 h isoproterenol’s chronotropic action is almost completely absent. All error bars represent mean and standard deviation. Different symbols represent experimental repeats across ten separate cell thaws. In general, suppressing HDACs leads to overall increase in beat rate and excitability while suppressing SIRTs leads to overall decrease in beat rate/excitability. B. The same data grouped by class of HDACs (I, IIa, IIb, III and IV). Statistical significance (p < 0.05) in difference from negative siRNA (panel A) or between groups (panel B), found by non-parametric Kruskal-Wallis test, is indicated by an (*). n=6–143 per group. C. Computational ionic model results quantifying changes in spontaneous rate in hiPSC-CMs using the Paci2020 model when decreasing or increasing IK1 and If separately and in tandem. Note…